KR20060134112A - Selection of ensemble averaging weights for a pulse oximeter based on signal quality metrics - Google Patents

Abstract

A method and a system for ensemble averaging signals in a pulse oximeter, including receiving first and second electromagnetic radiation signals from a blood perfused tissue portion corresponding to two different wavelengths of light, obtaining an assessment of the signal quality of the electromagnetic signals, selecting weights for an ensemble averager using the assessment of signal quality, and ensemble averaging the electromagnetic signals using the ensemble averager.

Description

Selection of Ensemble Averaging Weights Based on Signal Characteristic Metrics in Pulse Oxygen Saturation Meters

The present invention relates to an oxygen saturation meter, and more particularly, to the selection of an ensemble averaging weight for ensemble averaging signals comprising waveforms sensed by the pulse saturation meter.

Pulse oxygen saturation meters are used to measure various blood properties, including blood oxygen saturation and pulse rate of hemoglobin in the arterial blood of a patient. Measurement of these properties has been performed using a non-invasive sensor that scatters light through a portion of the tissue of the patient where blood flows and photoelectrically detects the absorption and diffusion of light in that tissue. The amount of light absorbed or diffused is used to measure the amount of blood components in the tissue using various algorithms known in the art. The "pulse" in the pulse saturation meter is measured by the amount of time-dependent change in arterial blood in the tissue during the heartbeat cycle. The signal processed by the sensed optical measurement is related to the periodic attenuation of optical energy through a portion of the tissue of the patient through which blood flows, and is similar to a plethysmographic waveform.

Ensemble averaging, a method of instantaneous averaging, is calculated using weights. In the pulse oxygen saturation meter, the ensemble averaging calculates the weighted averaging between the newly measured sample and the ensemble averaged sample before one pulse interval. The weights selected and used to calculate ensemble averaging have a significant impact on the ensemble averaging process. These weights may always be chosen equally or may be selected based on the characteristics of the signals to be ensemble averaged. For example, US Pat. No. 4,690,126 to Colon discloses an ensemble averaging where different weights are assigned to different pulses and composites, and the averaged pulse waveform is used to calculate oxygen saturation. Colon's signal metric for adjusting the ensemble averaging weight is based on measuring the degree of motion artifact, measuring the degree of low flow (eg pulse intensity below a certain limit), and pulse rate. .

On the other hand, there is a need for a more flexible and reliable method for selecting an ensemble averaging weight that is used to ensemble averaging signals including waveforms sensed by pulse saturation meters.

The present invention relates to the selection of an ensemble averaging weight that is used to ensemble average the signals associated with the sensed waveform in a pulse saturation meter. The selection of ensemble averaging weights is made based on one or more or a combination of various signal characteristic metrics or indicators. In one embodiment, the present invention provides a method of ensemble averaging signals in a pulse saturation concentration meter. The method includes receiving first and second electromagnetic radiation signals associated with two different wavelengths of light from tissue flowing blood; Obtaining an evaluation of a signal characteristic of the first and second electromagnetic signals; Selecting weights for use in an ensemble averaging using the evaluation of the signal characteristics; And ensembling the first and second electromagnetic signals using the ensemble averager.

In one embodiment, the selection of ensemble averaging weights includes evaluating and using various signal characteristic indicators, wherein the selection of weights comprises forming a combination of one or more signal characteristic parameters below. . The parameters may be determined by measuring the arrhythmia of the signals, measuring the similarity or interdependence between the first and second electromagnetic radiation signals, and measuring the degree of motion artifacts by the ratio of the long-term averaged pulse amplitude of the signal to the current pulse amplitude. The ratio of the current pulse amplitude to the previous pulse amplitude of the signal; And the ratio of the current pulse interval to the averaged pulse interval of the signal.

Features and advantages of embodiments of the present invention are described in more detail below with reference to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a block diagram showing an oxygen saturation meter as an example.

2 is a block diagram illustrating a signal processing structure of a pulse saturation measuring instrument according to an embodiment of the present invention.

A method and system according to one embodiment of the invention relates to the selection of an ensemble averaging weight that is used to ensemble average the signals associated with the waveform sensed by the pulse saturation meter. The selection of ensemble averaging weights is based on one or more or a combination of various signal characterizations or indicators. One embodiment of the present invention may be described and applied in connection with the pulse saturation rate and pulse or heart rate of hemoglobin in arterial blood, especially in the monitor of the pulse saturation meter and the sensor of the pulse saturation meter. . Meanwhile, the embodiment of the present invention may be equally applied to a general patient monitor related to ECG and blood pressure, and a patient sensor connected thereto, and may also be applied to a device other than an oxygen saturation meter or a pulse saturation meter.

A typical pulse oxygen saturation meter measures two physiological parameters, the oxygen saturation percentage (SpO 2 or sat) and pulse rate of arterial hemoglobin. Oxygen saturation can be measured by various techniques. In some techniques, the photocurrent generated by the light sensor is controlled and processed to determine the ratio of the red to infrared signals to the modulation ratio. It is known that this modulation rate is closely related to the oxygen saturation of arterial blood. Pulse oxygen saturation meters and sensors are empirically adjusted by measuring the range and modulation rate of oxygen saturation (SaO 2) measured in arterial blood of each population of patients, health examinees, or animals. This association can, on the contrary, be used to measure blood oxygen saturation (SpO 2) based on measurements of the patient's modulation rate. Determination of oxygen saturation using the modulation rate is described in U.S. Patent No. 5,853,364 issued on December 29, 1998, "METHOD AND APPARATUS FOR ESTIMATING PHYSIOLOGICAL PARAMETERS USING MODEL-BASED ADAPTIVE FILTERING" and issued March 27, 1990. Patent 4,911,167 describes " METHOD AND APPARATUS FOR DETECTING OPRICAL PULSES " which is referred to by the present invention. The relationship between oxygen saturation and modulation rate is described in US Pat. No. 5,645,059, “MEDICAL SENSOR WITH MODULATED ENCODING SCHEME,” issued July 8, 1997, and is referenced by the present invention. Most pulse saturation meters produce a pulse wave graph that includes the initially determined saturation or pulse rate that is susceptible to interference.

1 is a block diagram illustrating one embodiment of an oxygen saturation meter embodying the present invention. One embodiment of the present invention is implemented by a data processing algorithm executed by the microprocessor 122 described below. Light from the light source 110 passes through the inside of the tissue 112 through which blood flows, is scattered, and sensed by the light sensor 114. The sensor 100 comprising a light source and a light sensor further includes an encoder 116 that provides signals indicative of the wavelength of the light source 110 such that the oxygen saturation meter can select an appropriate calibration factor for calculating the oxygen saturation. can do. Encoder 116 may be a register, for example.

The sensor 100 is connected to the pulse oxygen saturation meter 120. This oxygen saturation meter includes a microprocessor 122 connected to an internal bus 124. Also connected to the bus is a RAM memory 126 and a display 128. The time processing unit (TPU) 130 provides a timing control signal to the light driving circuit 132 for controlling when the light source 110 emits light, and when a plurality of light sources are used, Provides multiplexed timing. The TPU 130 controls the gating-in of signals passing from the light sensor 114 to the amplifier 133 and the switching circuit 134. These signals are sampled at appropriate times based on the timing of light emission of the plurality of light sources. The received signal is passed through amplifier 136, low pass filter 136 and analog-to-digital converter 140. The digital data is stored in a serial module (QSM) 142 in a queue structure and then transferred to the RAM 126 when the QSM 142 becomes full. In one embodiment, there may be multiple parallel paths, with separate amplifiers, filters, and A / D converters, depending on the wavelength or spectra of the multiple lights received.

Based on the signal values received in relation to the light received at the light sensor 114, the microprocessor 122 calculates oxygen saturation using various algorithms. This algorithm requires coefficients that are related to the wavelength of light used and can be determined empirically. These coefficients are stored in ROM 146. In a two-wavelength system, the set of specific coefficients selected for a pair of wavelength spectra is determined by the value represented by the encoder 116 associated with the particular light source of the particular sensor 100. In one embodiment, multiple resistance values may be assigned to select a set of different coefficients. In another embodiment, the same resistors are used to select among the coefficients suitable for the near-red light source or the infrared light source paired with the far-red light source. The choice of whether to select a near red light set or a far red light set is made by a control input from the control inputs 154. The control inputs 154 may be, for example, a switch of a pulse saturation meter, a keyboard, or a port for providing commands from a remote host computer. In addition, a number of methods or algorithms can be used to determine the patient's pulse rate and oxygen saturation or other desired physiological parameters.

The above description of one embodiment of the pulse oxygen saturation measuring instrument is provided as basic information for explaining a method for adjusting an ensemble averaging weight applied to an ensemble averaging device, which will be described later with reference to FIG.

Embodiments of the present invention are implemented as part of a higher level signal processing system used to process optical signals having the purpose of operating a pulse saturation meter. Such a signal processing system is shown in FIG. 2 as a block diagram 200 of the signal processing structure of a pulse saturation meter according to one embodiment of the invention. The signal processing structure 200 according to an embodiment of the present invention may be implemented by a software algorithm executed by a processor of a pulse saturation meter. In addition, by adding oxygen saturation and pulse rate calculations, the system 200 can measure various signal metrics used to determine filter weight coefficients. Signal metrics indicate whether the pulse coincides with a pulse wave graph or with noise. Signal metrics may be related to, for example, frequency (in the range of human pulse rate), shape (in the form of heart pulse), rise time, and the like. The system shown in FIG. 2 calculates both oxygen saturation and pulse rate. System 200 is used to detect venous pulses and sensor off and pulse signal loss conditions, which will be described separately later.

I. Oxygen Saturation Calculation

Block 202 illustrates the operation of the signal adjustment block. The digitized red and infrared signals or waveforms are received at block 202, (1) first differentiated to remove baseline shift, (2) low pass filtered by a fixed coefficient, ( 3) Adjusted by dividing by the DC value to preserve the ratio. The function of the Signal Conditioning subsystem is to emphasize the high frequencies generated in the human pulse wave graph and to attenuate the low frequencies where the motion artifacts are generally concentrated. The signal conditioning subsystem selects its filter coefficients (broadband or narrowband) based on the characteristics of the hardware identified during initialization. The input to block 202 is digitized red light and infrared signals, the output of which is preprocessed red light and infrared signals.

Block 204 illustrates the operation of the pulse recognition and authentication block. Low pass filtered and digitized red and infrared signals are provided to the block to recognize the pulse and to verify that it is the same as the artery pulse. This is done using pre-trained neural nets, mainly by infrared signals. Pulse is recognized by measuring amplitude, shape and frequency. The input to block 204 is the average pulse interval from block 208. This function changes the preceding authentication using the pulse rate. The output of block 204 indicates the degree of arrhythmia and the characteristics of the individual pulses. The input to block 204 is (1) preprocessed red and infrared signals, (2) average pulse duration, and (3) lowpass waveform from the lowpass filter. The output from block 204 includes (1) the degree of arrhythmia, (2) the change in pulse amplitude, (3) the characteristics of the individual pulses, (4) the pulse tone notification, (5) the authorized pulse interval and age )to be.

Block 206 is used to calculate signal characteristic metrics. The block 206 determines the pulse shape (e.g., the slope by the derivative), the interval change, the amplitude and change of the pulse, the change in the ratio to the rate, and the frequency related to the pulse rate. Inputs to block 206 include (1) unprocessed digitized red light and infrared signals, (2) degree of arrhythmia, characteristics of individual pulses, changes in pulse amplitude, (3) preprocessed red light and infrared signals, (4) Average pulse interval. The output from block 206 is (1) low pass and ensemble averaging filter weights, (2) metrics for sensor off detectors, (3) normalized preprocessed waveforms, and (4) percent modulation.

Block 208 calculates an average pulse interval. The block 208 calculates an average pulse interval from the received pulses. Inputs to block 208 are the authenticated pulse interval and age. The output from block 208 includes an average pulse interval.

Block 210 represents the function of the low pass filter and the ensemble averaging subsystem. Block 210 low pass filtering and ensemble averaging the waveform normalized and preprocessed by block 206. The weight for the low pass filter is determined by the signal metrics of block 206. This signal is ensemble averaged by the ensemble averaging filter weight determined by the signal metrics of block 206 (by which other frequencies close to the pulse rate and their harmonics are reduced). Less weight is assigned when the signal is marked as being attenuated. More weight is assigned when the signal is labeled as irregular, because it is not suitable when the ensemble averaging is arrhythmia. Red light and infrared waveforms are processed separately, with the same filtering weights. The filtering is delayed by about 1 second so that the signal metrics can be calculated first.

The filters use successive variable weights. If the samples are not ensemble averaged, in weighted averaging, the weight for previously filtered samples is set to zero and new samples will be processed by this algorithm. The block tracks the cumulative amount of age and / or filtering of the signal (eg, the sum of response times and delays in processing). If a good pulse has not been detected for a certain period, too old results are labeled. Input to block 210 includes (1) normalized and preprocessed red and infrared signals, (2) average pulse interval, (3) low pass filter weight and ensemble averaging filter weight, and (4) ECG trigger if possible (5) reference infrared for a zero-crossing trigger. The output from block 210 is (1) filtered red and infrared signals, and (2) age.

Block 212 illustrates an operation of estimating a change in ratio relative to the ratio for the filtered waveform and calculating the averaging weight. The variable weight for the filtering is controlled by the change of the ratio to the ratio. Such filtering by variable weight has the effect of gently decreasing the ratio to the ratio as the artifact increases. The subsystem has two response modes, including quick mode and normal mode. For example, filtering in fast mode may target a 3 second age metric, and target age in normal mode may be 5 seconds. In quick mode, the current minimum weight is limited to the higher level. That is, the lower weight is assigned to the ratio calculation for the latest ratio when noise is present, and the upper weight is assigned when no noise is present. Inputs to block 212 are (1) filtered red and infrared signals and age, (2) calibration coefficients, and (3) response mode (eg, user speed setting). The output from block 212 is the averaging weight for the ratio calculation to the ratio. The averaging weight is used as input to block 214 along with the filtered infrared and red light waveforms for calculating the ratio and age for the averaged ratio.

Block 216 illustrates the operation of calculating oxygen saturation. Saturation is calculated by an algorithm that uses the correction factor and the ratio to the averaged ratio. Inputs to block 216 are (1) the ratio to the averaged ratio, and (2) the calibration coefficients. The output of block 216 is an oxygen saturation value.

II. Pulse rate calculation

Block 218 low pass filtering and ensemble averaging the signal (s) adjusted by block 202 to recognize the pulse rate. The weight for the low pass filter is determined by the signal metrics of block 206. The signal is also ensemble averaged by the ensemble averaging filter weight determined by the signal metrics of block 206 (by which other frequencies close to the pulse rate and their overtones are reduced). The response weight is assigned when the signal is marked as being attenuated. More weights are assigned when the signal is labeled as irregular because it is not suitable when the ensemble averaging is arrhythmia. Red light and infrared light are treated separately. The filtering is delayed for about 1 second so that the signal metrics can be calculated first.

The filters use continuous variable weights. If the samples will not be ensemble averaged, the weight for previously filtered samples in weighted averaging is set to zero and new samples are processed by this algorithm. Block 218 tracks the cumulative amount of age and / or filtering of the signal (eg, the sum of response times and delays in processing). Too old results are labeled (if a good pulse is not detected for a certain period of time). Input to block 218 includes (1) preprocessed red and infrared signals, (2) average pulse interval, (3) low pass filter weight and ensemble averaging filter weight, (4) ECG trigger, if possible ( 5) Reference infrared for zero-crossing trigger. The output of block 218 is (1) filtered red and infrared signals, and (2) age.

Block 220, or the filtered pulse recognition and authentication block, calculates a pulse interval from the filtered waveform, and this result is used only if the pulse has failed authentication by block 204. Inputs to block 220 are used to detect (1) filtered red and infrared signals and age, (2) average pulse duration, (3) hardware ID or noise rating, and (4) infrared and red light energy. Type or type of sensor. The output of block 220 includes an authenticated pulse interval and age.

Block 222, or pulse interval averaging and pulse rate calculation block, calculates the pulse velocity and averages the pulse intervals. Block 222 receives the authenticated pulse interval and age, and outputs (1) the average pulse interval, (2) pulse rate.

III. Pulsation of veins

Block 224, or venous pulsation sensing block, receives pulse rate and preprocessed red and infrared signals and age from block 202 and provides an indication of venous pulsation as output. Block 224 provides a reference infrared waveform in the time domain using a single-tooth comb filter that outputs to an ensemble averaging filter (eg, blocks 210 and 218). Inputs to block 224 are (1) filtered red and infrared signals and age, and (2) pulse rate. The output from block 224 is the marker of venous pulsation and the reference infrared. In one embodiment, block 224 plots the “openness” of the IR-Red Lissajous plot to determine whether a marker (eg, Venous_Pulsation) should be set. Measure The cover page output (eg Venous_Pulsation) is updated regularly (eg every second). The reference infrared waveform is also output to the ensemble averaging filter.

IV. Sensor off

Block 226, or the sensor off and loss of pulse amplitude detection block, utilizes pre-trained neural networks, for example, to determine whether the sensor is off at the surface of the tissue through which the patient's blood is flowing. Inputs to the neural network are metrics that quantify the response pattern of infrared and red light values over recent seconds. If the state of the signal does not indicate the presence of a pulse, or if the sensor does not indicate a viewing area (eg, pulse presence, disconnection, loss of pulse, sensor off estimation, and sensor off), the samples may be stored in the system 200. In many subsystems). The input to block 226 is an ID indicating (1) signal characteristic metrics, (2) LED brightness of the oxygen saturation meter, amplifier gain, and (3) hardware configuration of the oxygen saturation meter. The output from block 226 includes the state of the signal including an indication for sensor off.

In the structure 200 described above, the pulse loss and pulse search beacons, which are functions of block 226, can be derived using information from several portions of the signal processing structure. On the other hand, if a suitable sensor is not connected or if a sensor off or loss of pulse amplitude is detected by the signal processing structure, the signal processing structure does not use the received infrared and red light waveforms to calculate oxygen saturation or pulse rate. Will not.

The description of the pulse oxygen saturation meter signal processing structure according to the present invention as described above is ensemble used to ensemble the signals associated with the waveforms detected from the pulse oxygen saturation meter, represented by blocks 210 and 218. It is provided as a basis for describing a method and apparatus for selecting an averaging weight.

Ensemble Averaging Weights

As noted above, the selection of ensemble averaging weights is made based on one, many or a combination of various signal characteristic metrics or indicia. In particular, in one embodiment, the metrics used to adjust the ensemble averaging weight include a measure of the degree of arrhythmia. Since ensemble averaging is inadequate for signals with rapidly changing frequencies, this metric is used to reduce the degree of ensemble averaging when arrhythmias appear in patients. Another metric used to adjust the ensemble averaging weights includes a measure of the degree of change of the ratio to the ratio (eg, lack of similarity or relevance between the infrared and red light waveforms). This metric is sensitive to the presence of motion or other sources of noise. Unlike the known Conlon's method, which suggests a metric that probably compares similarities between current and previous pulse waveforms obtained from the same waveform, this metric uses similarity between two identical waveforms with different wavelengths. . Another metric used to adjust the ensemble averaging weight includes the ratio of the current pulse amplitude to the long term average pulse amplitude. The long-term average pulse amplitude is related to the average with a response time of about one minute when all pulses are authenticated, and the time may be increased if most pulses can be authenticated. Similar to Conlon's approach, this metric is for detecting motion artifacts, but this metric is an analog metric (Conlon's approach has a few discrete states such as non-artifacts, low artifacts, and high artifacts). Another metric used to adjust the ensemble averaging weight includes the ratio of the current pulse amplitude to the previous pulse amplitude. This metric is used to quickly change the ensemble averaging weight when large motion artifacts are started or stopped. Another metric used to adjust the ensemble averaging weight includes the measurement of all signal characteristic metrics for a single pulse, which is a combination with itself and several other metrics including the aforementioned metrics. Use this metric to quickly reduce ensemble filtering when motion artifacts are calmed and the characteristics of the input waveform are estimated to be improved over heavily ensemble averaged waveforms. Another metric used to adjust the ensemble averaging weight includes the ratio of the current pulse interval to the average pulse interval. Use this metric to reduce ensemble filtering in cases where the heart skips pulsations, which can sometimes occur in many people.

If the subsystem 210 and / or 218 has recognized that the pulse recognition and authentication subsystem 204 has completed the evaluation of the potential pulse, the subsystem may be used to ensemble averaging used for an instance of the ensemble averaging subsystem. Update the weights. To calculate two ensemble averaging instances with outputs used to calculate saturation and pulse rate, the respective weights are calculated. These weights are based on the metrics provided by the instance of the pulse recognition and authentication subsystem having an input waveform that is not ensemble averaged.

The calculation for Sat_Ensemble_Averaging_Filter_Weight is as follows:

x = max (Short_RoR_Variance, Pulse_Qual_RoR_Variance / 1.5) * max (Long_Term_Pulse_Amp_Ratio, 1.0)

RoR_Variance_Based_Filt_Wt = 0.5 * 0.05 / max (0.05, x)

Arr_Prob = (Period_Var-0.1 * Short_RoR_Variance-0.09) / (0.25-0.09);

Arr_Min_Filt_Wt_For_Sat = 0.05 + 0.5 * bound (Arr_Prob, 0, 1.0)

Sat_Ensemble_Averaging_Filter_Weight = max (RoR_Variance_Based_Filt_Wt, Arr_Min_Filt_Wt_For_Sat) * (1.0 + Pluse_Qual_Score)

Sat_Ensemble_Averaging_Filter_Weight = min (Sat_Ensemble_Averaging_Filter_Weight, 1.0)

Here, bound (a, b, c) means min (max (a, b), c).

The formula generates a default weight of 0.5 for the lower value of the rate change relative to the rate. Short_RoR_Variance and Pulse_Qual_RoR_Variance are calculated for a time of 3 seconds, for example. The interval for Pulse_Qual_RoR_Variance is terminated by authentication and rejection of the latest pulse expected to contain the latest sample. The weight is reduced by a high change in the ratio to the ratio, or by a high value of Long_Term_Pulse_Amp_Ratio that represents the motion artifact. Arr_Min_Filt_Wt_For_Sat gives a minimum value of the ensemble averaging weights (range 0.05-0.55) based on Period_Var indicating the degree of arrhythmia. Since ensemble averaging is not effective for pulses with different intervals, this operation is performed. If the latest pulse has a good Pulse_Qual_Score, the maximum value of Sat_Ensemble_Averaging_Filter_Weight can be increased from 0.5 to 1.0.

The formula for Rate_Ensemble_Averaging_Filter_Weight is:

Arr_Prob = (Period_Var-0.07) / (0.20-0.07)

Arr_Min_Filt_Wt_For_Rate = 0.05 + 0.5 * bound (Arr_Prob, 0, 1.0)

x = max (RoR_Variance_Based_Filt_Wt, Arr_Min_Filt_Wt_For_Rate) * (1.0 + Pulse_Qual_Score)

if Short_Term_Pulse_Amp_Ratio * Long_Term_Pulse_Amp_Ratio <1.0

x = x / Short_Term_Pulse_Amp_Ratio

if Avg_Period> 0

x = x * bound (Pulse_Qual_Score * Qualified_Pulse_Period / Avg_Period, 1.0, 3.0)

Rate_Ensemble_Averaging_Filter_Weight = min (x, 1.0)

This formula differs from the Sat_Ensemble_Averaging_Filter_Weight formula as follows:

a) The thresholds used to calculate Arr_Prob are rather low, since it is desirable to ensemble averaging prior to pulse authentication to clarify the irregular pulse.

b) Small values of Short_Term_Pulse_Amp_Ratio indicate whether the motion artifact is calm and mean that the ensemble averaging weight increases rapidly. This has been found to be useful for pulse authentication, but it is not useful for filtering ratios and saturation calculations for ratios.

c) If the heart skips the pulsation, even before or without arrhythmias before, and Qualified_Pulse_Period is longer than average, it will increase the ensemble averaging weight to clarify the pulses skipped by subsequent pulse authentications. will be.

In one embodiment, in two separate ensemble averaging machines for processing the sensed waveforms used to calculate the oxygen saturation and pulse rate, the ensemble averaging weights determined as described above are used. The ensemble averaging device used to calculate the oxygen saturation is operated by a normalized signal, and the ensemble averaging device for pulse rate calculation is operated by an unnormalized signal. Pulse saturation meters that individually ensemble the oxygen saturation and pulse rate individually are described in pending patents, "Pulse Oximeter with Separate Ensemble Averaging for Oxygen Saturation and Heart Rate, Attorney Docket: TTC-009103-022700US" The contents are referred to in the present invention. In the patent application, the metrics selected for the two paths towards the two ensemble averaging may be altered to optimize the ensemble averaging for the calculation of oxygen saturation or pulse rate. For example, a lower threshold may be used for a metric for detecting an irregular pulse during calculating the pulse rate, as compared to calculating the oxygen saturation. In addition, as motion artifacts subside, the metric for short-term pulse amplitude ratios becomes smaller, which will be given more weight in pulse rate calculations than in the calculation of oxygen saturation.

Definition of terms:

(Data entry)

Avg_Period-Average pulse interval provided by the pulse rate calculation subsystem.

Long_Term_Pulse_Amp_Ratio-The ratio of the recent pulse amplitude to the past pulse amplitude. Provided by Pulse Recognition and Authentication Subsystem. In practice, values greater than 1.0 are recognized as motion artifacts, resulting in a reduction of Ensemble_Averaging_Filter_Weights.

Period_Var-Period change metric from the pulse recognition and authentication subsystem. Used to measure the range of arrhythmia. For example, a value of 0.10 indicates that a series of pulse intervals vary within 10% of the average in Avg_Period.

Pulse_Qual_RoR_Variance-RoR_Variance metric from pulse recognition and authentication subsystem.

Pulse_Qual_Score-Score calculated by the Pulse Recognition Neural Network of the Pulse Recognition and Authentication Subsystem. 0 is the lowest score, 1.0 is the highest score.

Qualified_Pulse_Period-The latest pulse interval certified by the Pulse Recognition and Authentication Subsystem.

Short_Term_Pulse_Amp_Ratio-The ratio of the latest pulse amplitude to the previous pulse amplitude.

(Print)

Frequency_Ratio-The ratio of Mean_IR_Frequency_Content to pulse rate.

LPF_RoR_Variance-Quantify the rate of change of the rate against the rate. Calculated for 9 seconds window from LPF_Scaled_Waveforms.

Rate_LPF_Weight-Low pass filter weight used by an instance of the ensemble averaging subsystem that preprocesses the waveforms used to calculate pulse characteristics and pulse rate.

RoR_Variance-Quantify the amount of change in the ratio to the ratio. Calculated during the 9 second window from Scaled_Waveforms. For example, a value of 0.10 indicates that the ratio value for the ratio of sample to sample varies in the range of 10% with respect to the mean value of the ratio for ratio.

Sat_Ensemble_Averaging_Filter_Weight-Ensemble averaging weight used by an instance of the ensemble averaging subsystem that preprocesses the waveforms used for pulse authentication and pulse rate calculation.

Sat_LPF_Weight-Low pass filter weight used by an instance of the ensemble averaging subsystem that preprocesses the waveform used for pulse authentication and pulse rate calculation.

Scaled_Waveforms-A scaled version of Pre_Processed_Waveforms for infrared and red light.

Short_RoR_Variance-Quantify the rate of change of the rate against the rate. Calculated for 3 seconds window from Scaled_Waveforms

(Internal variables)

Arr_Prob-Possibility of arrhythmia to limit the amount of ensemble averaging. Based on the threshold specified for each of the two Ensemble_Averaging_Filter_Weights, and Period_Var.

Arr_Min_Filt_Wt_For_Rate, Arr_Min_Filt_Wt_For_Sat-The minimum of two Ensemble_Averaging_Filter_Weights based on each Arr_Prob value.

LPF_Scaled_Waveforms-Low pass filtered version of Scaled_Waveforms used to calculate LPF_RoR_Variance.

Mean_IR_Frequency_Content-Estimation of the average frequency of the infrared input waveform. Used to calculate the Frequency_Ratio metric.

RoR_Variance_Based_Filt_Wt-Component of Ensemble_Averaging_Filter_Weights based on RoR_Variance metrics and Long_Term_Pulse_Amp_Ratio.

The invention relating to the selection of ensemble averaging weights may have various embodiments without departing from the basic features of the invention as will be appreciated by those skilled in the art. For example, although the present invention uses the time-domain, frequency-based methods can be equally applied. Accordingly, the foregoing descriptions are given by reference only, and the scope of the present invention should be defined by the appended claims.

Claims (17)

Receiving first and second electromagnetic radiation signals corresponding to two different wavelengths of light from the blood flowing tissue; Obtaining an evaluation of a signal characteristic of the first and second electromagnetic signals; Selecting weights for use in an ensemble averaging using the evaluation of the signal characteristics; And Ensembling the first and second electromagnetic signals using the ensemble averager; Method for ensemble averaging signals in an oxygen saturation meter comprising a. The method of claim 1, Obtaining an evaluation of the signal characteristic comprises obtaining a measurement of the degree of arrhythmia of the signals. The method of claim 2, Acquiring an evaluation of the signal characteristic further comprises obtaining a measurement of a degree of similarity or association between the first and second electromagnetic radiation signals to ensemble the signals in an oxygen saturation meter. Way. The method of claim 1, Acquiring an evaluation of the signal characteristic comprises obtaining a measurement of the degree of motion artifact present in the signals. The method of claim 4, wherein Acquiring the measure of the degree of motion artifact comprises obtaining a ratio of the current pulse amplitude to the long-term average pulse amplitude of the signal. The method of claim 1, Acquiring an evaluation of the signal characteristic comprises obtaining a ratio of a current pulse amplitude to a previous pulse amplitude of the signal. The method of claim 1, Obtaining an assessment of the signal characteristics comprises averaging the signals in an oxygen saturation meter comprising obtaining measurements of the magnitude of all signal characteristic metrics for a single pulse, including combinations with other metrics. Way. The method of claim 1, Obtaining an evaluation of the signal characteristic comprises obtaining a ratio of a current pulse interval to an average pulse interval of the signals. The method of claim 1, Selecting the weights may include measuring an irregular level of the signals, a measure of the degree of similarity and relevance between the first and second electromagnetic radiation signals, and a current for long-term averaged pulse amplitudes of the signals. One or more parameters selected from the acquisition of the ratio of pulse amplitudes, the measurement of the degree of motion artifacts by the ratio of the current pulse amplitude to the previous pulse amplitude of the signal and the ratio of the current pulse interval to the averaged pulse interval of the signals. And ensemble averaging signals in an oxygen saturation meter comprising forming a combination of the two. Means for receiving first and second electromagnetic radiation signals associated with two different wavelengths of light from the blood flowing tissue; Means for obtaining an evaluation of a signal characteristic of the first and second electromagnetic signals; Means for selecting a weight for use in an ensemble averaging using the evaluation of the signal characteristics; And An ensemble averaging unit that ensembles the first and second electromagnetic signals using the weights; Apparatus for ensembling the signal in the oxygen saturation meter comprising a. The method of claim 10, And means for obtaining an assessment of the signal characteristics is to obtain a measurement of the degree of arrhythmia of the signals. The method of claim 11, And means for obtaining an evaluation of the signal characteristic obtains a measurement of the degree of similarity or association between the first and second electromagnetic radiation signals. The apparatus for ensemble averaging signals in an oxygen saturation meter. . The method of claim 10, And the means for obtaining an assessment of the signal characteristics obtains a measurement of the degree of motion artifacts present in the signals. The method of claim 10, And means for obtaining an assessment of the signal characteristics obtains a ratio of the current pulse amplitude to the long term average pulse amplitude of the signal. The method of claim 10, And means for obtaining an evaluation of the signal characteristic obtains a ratio of a current pulse amplitude to a previous pulse amplitude of the signal. The method of claim 10, The means for obtaining an assessment of the signal characteristics comprises obtaining a measurement of the degree of all signal characteristic metrics for a single pulse, including combinations with other metrics, to ensemble the signals in an oxygen saturation meter. Device. The method of claim 10, And means for obtaining an evaluation of the signal characteristic obtains a ratio of the current pulse interval to the average pulse interval of the signals.

Images